There are many unproven methods of ventilation and oxygenation that have been used in children dying from respiratory failure, including extracorporeal devices and high frequency ventilation. However, no objective predictors of death have been developed that enable clinicians to decide when it is reasonable to use these therapies in children over one month of age with acute respiratory failure. To determine predictors of death, we evaluated the charts of all 42 children aged one month to eighteen years admitted to the intensive care unit who were ventilated for more than twelve hours, received greater than 90% oxygen, received a peak inspiratory pressure greater than 25 cmH2O and had no pre-existing neurodevelopmental handicap. A combination of a variable reflecting ventilation and a variable reflecting oxygenation reliably predicted death: a combination of ventilation index greater than 40 and oxygenation index greater than 0.4 was associated with a 77% chance of mortality (sensitivity 65% and specificity 74%); a combination of peak inspiratory pressure greater than 40 cmH2O and A-aDO2 greater than 580 was associated with an 81% chance of mortality (sensitivity 74% and specificity 79%). These criteria may be useful in predicting death from respiratory failure in children.